Introduction

Research Question

What is STAG2? What should we expect if we knocked out STAG2 from a cell?

STAG2 is a important subunit in the cohesin complex that plays an important role in regulating sister chromatid alignment during cell division, and other genetic regulatory functions. Studies show that STAG2-mut EWS have higher rates of metastatic disease and worse outcomes. It is expected that the TC71 & A673 cell lines will have differing phenotype effects to STAG2 KO, for example STAG2 deletion will lead to TC71 growth defect but with A673 it will lead to a growth advantage. STAG1 levels could also increase, in order to possibly fill in for the removal of STAG2 in the cell cycle, a decrease of RAD21 is also observed. The cohesin complex also mediates intrachromosomal interactions including those conjoining enhancers to promoters. Loss of STAG2 produces highly consistent and stable transcriptional changes that may undergo selection to confer a competitive advantage. Two neurodvelopmental transcription factors, POU3F2 and NR2F1 were consistently upregulated in STAG2 KO studies. (Adane et al, 2022)

STAG2 and CDKN24 seem to share an exclusive pattern of genetic alterations. cell lines with STAG2 mutations seem likely to express p16 , and reciprocally all cases with CDKN2A deletion seem to express STAG2. (Tirode et al. 2015)

In a study by CellPress, a GSEA enrichment analysis was conducted and from the 18,889 signatures ranked by the average normalized enrichment score (NES), several of the top 20 signatures enriched in STAG2 proficient condition were EWSR1-FLI1-regulated gene signatures. (Surdez et al. 2021)

The expected pathways for Ewing Sarcoma was also reviewed to compare with the gsea and enrichr results. For a study by JBUON, multiple test were conducted first we’ll list is the top 10 significantly up-regulated and down-regulated top: UGT3A2, HMCN1, RBM11, DKK2, SNORA23, PTPN13, TNFAIP6, LIPI, DCC, HOXD10 and down: ATP1B1, CLU, MAOB, SORBS1, SORL1, SYNPO2, KIAA1324, GATM, IGKV2D-28, CKMT1B. and for the GO enrichment analysis, for BP ontology pathways relating to transcription, chromatin modfication and remodeling, SRP-dependent cotranslational protein targeting to membrane, viral transcription, rRNA processing, DNA replication. for the CC enriched pathways, nucleus, nuceloplasm, nucleous, centrosome, cytoplasm, nuclear speck, focal adhesion, nuclear membrane, membrane, nuclear chromatin. For MF ontology the pathways are poly(A) RNA binding, DNA binding, chromatin binding, nucleic acid binding, protein binding, nucleotide binding, helicase binding. (Yan et al. 2018)

Data input

All samples are downloaded from the SRA Run Selector BioProject PRJNA549593, slight quality control via fastp was conducted and then the reads were aligned using Salmon for Transcript-level quantification files. The metadata for the samples was provided from the SRA Run Selector also. 3 sample comparisons were constructed using the sample data:

  1. SA2 KO vs WT in A673 cells.

  2. SA2 KO vs WT in TC71 cells.

samples <- read.table(file.path("Analysis/SraRunTable.txt"), sep = ",", header = TRUE) %>%
  dplyr::mutate(cell_line = ifelse(grepl("A673", x = source_name), "A673", "TC71")) %>%
  dplyr::mutate(condition = ifelse(grepl("WT|siCT", x = GENOTYPE), "Control", "Treatment"))

A673_samples <- samples %>%
  dplyr::filter(GENOTYPE %in% c("WT", "SA2 KO") & cell_line == "A673")

TC71_samples <- samples %>%
  dplyr::filter(GENOTYPE %in% c("SA2 KO", "WT") & cell_line == "TC71")

A673_salmon_files <- file.path("Salmon/salmon.out", A673_samples$Run, "quant.sf") %>%
  setNames(object = , A673_samples$Run)

TC71_salmon_files <- file.path("Salmon/salmon.out", TC71_samples$Run, "quant.sf") %>%
  setNames(object = , TC71_samples$Run)

ensdb <- EnsDb.Hsapiens.v86

transcripts <- transcripts(ensdb, columns = c(listColumns(ensdb, "tx"), "gene_name"), return.type = "data.frame") %>%
  as_tibble() %>%
  dplyr::select(tx_id, gene_name)

A673_txi <- tximport(A673_salmon_files, type = "salmon", tx2gene = transcripts, ignoreTxVersion = TRUE)

TC71_txi <- tximport(TC71_salmon_files, type = "salmon", tx2gene = transcripts, ignoreTxVersion = TRUE)

A673_dds_txi <- DESeqDataSetFromTximport(A673_txi, colData = A673_samples, design = ~condition)

TC71_dds_txi <- DESeqDataSetFromTximport(TC71_txi, colData = TC71_samples, design = ~condition)

Collected Sample Metadata

data_input_all <- samples %>%
  dplyr::select(GENOTYPE, cell_line) %>%  
  kbl(caption = "Table 1: Collected Sample Overview") %>%
  kable_styling(bootstrap_options = "striped", full_width = T, html_font = "Cambria")
data_input_all
Table 1: Collected Sample Overview
GENOTYPE cell_line
SA1 KO A673
SA1 KO A673
SA1 KO A673
SA2 KO A673
SA2 KO A673
SA2 KO A673
Rescue of SA2 in SA2 KO cell A673
Rescue of SA2 in SA2 KO cell A673
Rescue of SA2 in SA2 KO cell A673
siCT A673
siCT A673
siEF1 A673
siEF1 A673
WT A673
WT A673
WT A673
SA2 KO TC71
SA2 KO TC71
SA2 KO TC71
SA2 KO TC71
SA2 KO TC71
SA2 KO TC71
siCT TC71
siCT TC71
siEF1 TC71
siEF1 TC71
WT TC71
WT TC71
WT TC71

Exploratory Data

PCA Plots

vst <- vst(A673_dds_txi)

A673_PCA <- plotPCA(vst, intgroup = c("cell_line", "GENOTYPE"), returnData = TRUE)
A673_percentVar <- round(100 * attr(A673_PCA, "percentVar"))

ggplot(A673_PCA, aes(PC1, PC2, color = GENOTYPE)) +
  geom_point(size = 3) +
  ggtitle("PCA Plot for A673 STAG2KO samples") +
  xlab(paste0("PC1: ", A673_percentVar[1], "% variance")) +
  ylab(paste0("PC2: ", A673_percentVar[2], "% variance")) +
  coord_fixed()

TC71_vst <- vst(TC71_dds_txi)

TC71_PCA <- plotPCA(TC71_vst, intgroup = c("cell_line", "GENOTYPE"), returnData = TRUE)
TC71_percentVar <- round(100 * attr(TC71_PCA, "percentVar"))

ggplot(TC71_PCA, aes(PC1, PC2, color = GENOTYPE)) +
  geom_point(size = 3) +
  ggtitle("PCA Plot for TC71 STAG2KO samples") +
  xlab(paste0("PC1: ", TC71_percentVar[1], "% variance")) +
  ylab(paste0("PC2: ", TC71_percentVar[2], "% variance")) +
  coord_fixed()

Analysis Samples Metadata

A673_metadata <- colData(A673_dds_txi)
TC71_metadata <- colData(TC71_dds_txi)

sample_metadata <- rbind(A673_metadata, TC71_metadata)

sample_metadata %>%
  as.data.frame() %>%
  dplyr::select(GENOTYPE, cell_line) %>%
  kbl(caption = "Table 2: Analysis Sample Overview") %>%
  kable_styling(bootstrap_options = "striped", full_width = T, html_font = "Cambria")
Table 2: Analysis Sample Overview
GENOTYPE cell_line
SRR9326185 SA2 KO A673
SRR9326186 SA2 KO A673
SRR9326187 SA2 KO A673
SRR9326195 WT A673
SRR9326196 WT A673
SRR9326197 WT A673
SRR9326198 SA2 KO TC71
SRR9326199 SA2 KO TC71
SRR9326200 SA2 KO TC71
SRR9326201 SA2 KO TC71
SRR9326202 SA2 KO TC71
SRR9326203 SA2 KO TC71
SRR9326208 WT TC71
SRR9326209 WT TC71
SRR9326210 WT TC71

Results

Statistical Analysis

Volcano Plots

EnhancedVolcano(A673_sig_ordered_result,
  lab = A673_sig_ordered_result$Gene_name,
  x = "log2FoldChange",
  y = "pvalue",
  title = "Siginifcant Genes for STAG2KO in A673 cells",
  subtitle = "",
  pointSize = 1.0,
  labSize = 4.0,
  xlim = c(min(A673_sig_ordered_result$log2FoldChange), max(A673_sig_ordered_result$log2FoldChange)),
  ylim = c(0, 300)
)

EnhancedVolcano(TC71_sig_ordered_result,
  lab = TC71_sig_ordered_result$Gene_name,
  x = "log2FoldChange",
  y = "pvalue",
  title = "Siginifcant Genes for STAG2KO in TC71 Cells",
  subtitle = "",
  pointSize = 1.0,
  labSize = 4.0,
  xlim = c(min(TC71_sig_ordered_result$log2FoldChange), max(TC71_sig_ordered_result$log2FoldChange)),
  ylim = c(0, 200)
)

Significant DEG tables

A673_table_result <- dplyr::select(A673_sig_ordered_result, Gene_name, log2FoldChange, stat, pvalue, padj)

datatable(A673_table_result,
  class = "cell-border stripe",
  caption = "Table 3: A672 STAG2KO Differentally Significant Genes", rownames = FALSE
)
TC71_table_result <- dplyr::select(TC71_sig_ordered_result, Gene_name, log2FoldChange, stat, pvalue, padj)

datatable(TC71_table_result,
  class = "cell-border stripe",
  caption = "Table 4: STAG2KO samples Differentally Significant Genes", rownames = FALSE
)

STAG2 Plot Counts

A673_counts <- plotCounts(A673_dds, gene = "STAG2", returnData = TRUE)

ggplot(A673_counts, aes(x =condition, y = count, color = condition)) +
  geom_point(size = 2) +
  ggtitle("Count of STAG2 for A672 SA2KO vs WT")

TC71_counts <- plotCounts(TC71_dds, gene = "STAG2", returnData = TRUE)

ggplot(TC71_counts, aes(x =condition, y = count, color = condition)) +
  geom_point(size = 2) +
  ggtitle("Count of STAG2 for TC71 SA2KO vs WT")

Venn Diagram

A673_overexpressed <- A673_sig_ordered_result %>%
  dplyr::filter(log2FoldChange > 0)
  
A673_underexpressed <- A673_sig_ordered_result %>%
  dplyr::filter(log2FoldChange < 0) 
  
TC71_overexpressed <- TC71_sig_ordered_result %>%
  dplyr::filter(log2FoldChange > 0) 
  
TC71_underexpressed <- TC71_sig_ordered_result %>%
  dplyr::filter(log2FoldChange < 0) 
  
grid.newpage()

over <- venn.diagram(list(A673_overexpressed$Gene_name, TC71_overexpressed$Gene_name),
                     category.names = c("A673","TC71"),
                     filename = NULL,
                     main = "Overlapping Overexpressed Genes",
                     fill = c("red", "deepskyblue4"), 
                     lwd = 1, lty = 1)
grid.draw(over)

grid.newpage()

under <- venn.diagram(list(A673_underexpressed$Gene_name, TC71_underexpressed$Gene_name),
                     category.names = c("A673","TC71"),
                     filename = NULL,
                     main = "Overlapping Underexpressed Genes",
                     fill = c("red", "deepskyblue4"), 
                     lwd = 1, lty = 1)
grid.draw(under)

4-way plot

A673_list <- A673_result_df %>%
  dplyr::select(log2FoldChange, stat, padj) %>%
  rownames_to_column() %>%
  dplyr::rename("Gene" = rowname, "A673_Log2FC" = log2FoldChange, "A673_stat" = stat, "A673_padj" = padj) %>%
  drop_na()


TC71_list <- TC71_result_df %>%
  dplyr::select(log2FoldChange, stat, padj) %>%
  rownames_to_column() %>%
  dplyr::rename("Gene" = rowname, "TC71_Log2FC" = log2FoldChange, "TC71_stat" = stat, "TC71_padj" = padj) %>%
  drop_na()
  
fourway_df <- inner_join(A673_list, TC71_list, by = "Gene") %>%
  dplyr::mutate(Sig_Group = case_when(
    A673_padj < 0.05 & TC71_padj < 0.05 ~ "Both", 
    A673_padj < 0.05 ~ "A673-only", TC71_padj < 0.05 ~ "TC71-only",
    TRUE ~ "Not Significant"
  ))
  
dge_fourway <- ggplot(data = fourway_df, aes(x = A673_stat, y = TC71_stat, label = Gene, color = Sig_Group)) +
  geom_point(alpha = .8) +
  geom_hline(yintercept = 0, size = .1) +
  geom_vline(xintercept = 0, size = .1) +
  coord_fixed(ratio = 1) +
  scale_color_manual(values = c(
    "Both" = "blue",
    "A673-only" = "darkgoldenrod2",
    "TC71-only" = "firebrick",
    "Not Significant" = "grey")) +
  labs(title = "Wald stat of Differentialy Expressed Genes in A673 and TC71 Cell lines")

ggplotly(dge_fourway)

Top 10 Over-expressed and 10 Under-expressed heatmaps

A673_heatmap <- pheatmap(A673_sig_norm_dds_counts,
  main = "Top 10 Over- and Under- expressed A673 STAG2KO DEGs",
  color = palette(200),
  cluster_rows = FALSE,
  cluster_cols = FALSE,
  show_rownames = TRUE,
  annotation = dplyr::select(A673_heat_meta, condition),
  scale = "row"
)

TC71_heatmap <- pheatmap(TC71_sig_norm_dds_counts,
  main = "Top 10 Over- and Under- expressed TC71 STAG2KO DEGs",
  color = palette(200),
  cluster_rows = FALSE,
  cluster_cols = FALSE,
  show_rownames = TRUE,
  annotation = dplyr::select(TC71_heat_meta, condition),
  scale = "row"
)

GSEA Enrichment Analysis

GSEA Dot Plot

dotplot(A673_gsea_res, font.size = 8) +
  ggtitle("A673 GSEA")

dotplot(TC71_gsea_res, font.size = 8) +
  ggtitle("TC71 GSEA")

GSEA Pathway Plots

A673_gsea_df <- as.data.frame(A673_gsea_res)

A673_top_pathways <- A673_gsea_df %>%
  top_n(n = 4, wt = NES) %>%
  pull(ID)
A673_top_pathways_plot <- gseaplot2(A673_gsea_res, geneSetID = A673_top_pathways, title = "Top 4 top A673 Pathways") 

A673_top_pathways_plot

A673_bot_pathways <- A673_gsea_df %>%
  top_n(n = 4, wt = -NES) %>%
  pull(ID)
A673_bot_pathways_plot <- gseaplot2(A673_gsea_res, geneSetID = A673_bot_pathways, title = "Top 4 bottom A673 Pathways") 

A673_bot_pathways_plot

TC71_gsea_df <- as.data.frame(TC71_gsea_res)

TC71_top_pathways <- TC71_gsea_df %>%
  top_n(n = 4, wt = NES) %>%
  pull(ID)
TC71_top_pathways_plot <- gseaplot2(TC71_gsea_res, geneSetID = TC71_top_pathways, title = "Top 4 top TC71 Pathways") 

TC71_top_pathways_plot

TC71_bot_pathways <- TC71_gsea_df %>%
  top_n(n = 4, wt = -NES) %>%
  pull(ID)
TC71_bot_pathways_plot <- gseaplot2(TC71_gsea_res, geneSetID = TC71_bot_pathways, title = "Top 4 bottom TC71 Pathways") 

TC71_bot_pathways_plot

GSEA Venn Diagram

pathways <- ggVennDiagram(x = list(A673 = A673_gsea_df$ID, TC71 = TC71_gsea_df$ID)) +
  ggtitle("Overlapping Pathways between Cell Lines") +
  theme(plot.title = element_text(hjust = 0.5)) +
  scale_color_brewer(palette = "Blues") +
  scale_fill_gradient(low = "#ece7f2", high = "#2b8cbe")
pathways

GSEA Pathway Datatables

path_overlap <- intersect(A673_gsea_df$ID, TC71_gsea_df$ID) 
"%!in%" <- Negate("%in%")

cell_gsea_list <- list('A673' = A673_gsea_df, "TC71" = TC71_gsea_df)

pathway_data <- lapply(cell_gsea_list, function(x){ (x[x$ID %!in% path_overlap,] %>%
                                                       rownames_to_column("Pathway") %>%
                                                       dplyr::arrange(rank) %>%
                                                       dplyr::select(c(-ID, -Description, -setSize,
                                                                       -enrichmentScore, -qvalue, -leading_edge,
                                                                       -core_enrichment)) %>%
                                                       dplyr::arrange(desc(NES)))
  })

pathway_data$A673 %>%
  datatable(caption = "Table 5: A673 GSEA Pathways")
pathway_data$TC71 %>%
  datatable(caption = "Table 6: TC71 GSEA Pathways")
both_pathways <- inner_join(A673_gsea_df, TC71_gsea_df, by = "ID") %>%
  dplyr::mutate(meanrank = rowMeans(cbind(rank.x, rank.y))) %>%
  dplyr::arrange(meanrank) %>%
  dplyr::select(starts_with(c("ID", "NES", "p.adjust")))

names(both_pathways) <- gsub("*.y", ".TC71", names(both_pathways))
names(both_pathways) <- gsub("*.x", ".A673", names(both_pathways))

datatable(both_pathways, caption = "Table 7: Shared Pathways for Both Cell Lines")

Enrichr Pathway Analysis

A673 enrichr Pathway enrichment

A673_resRmd <- llply(names(A673_gene_list), function(groupNow) {
  genesNow <- A673_gene_list[[groupNow]]
  response <- httr::POST(
    url = "https://maayanlab.cloud/Enrichr/addList",
    body = list(
      "list" = paste0(genesNow, collapse = "\n"),
      "description" = groupNow
    )
  )
  response <- jsonlite::fromJSON(httr::content(response, as = "text"))
  permalink <- paste0(
    "https://maayanlab.cloud/Enrichr/enrich?dataset=",
    response$shortId[1]
  )
  knitr::knit_child(
    text = c(
      "#### `r groupNow`",
      "",
      'Enrichr Link: <a href="`r permalink`" target="_blank">`r groupNow`</a>.',
      ""
    ),
    envir = environment(),
    quiet = TRUE
  )
})
cat(unlist(A673_resRmd), sep = "\n")

Over-expressed

Enrichr Link: Over-expressed.

Under-expressed

Enrichr Link: Under-expressed.

TC71 enrichr Pathway enrichment

TC71_resRmd <- llply(names(TC71_gene_list), function(groupNow) {
  genesNow <- TC71_gene_list[[groupNow]]
  response <- httr::POST(
    url = "https://maayanlab.cloud/Enrichr/addList",
    body = list(
      "list" = paste0(genesNow, collapse = "\n"),
      "description" = groupNow
    )
  )
  response <- jsonlite::fromJSON(httr::content(response, as = "text"))
  permalink <- paste0(
    "https://maayanlab.cloud/Enrichr/enrich?dataset=",
    response$shortId[1]
  )
  knitr::knit_child(
    text = c(
      "#### `r groupNow`",
      "",
      'Enrichr Link: <a href="`r permalink`" target="_blank">`r groupNow`</a>.',
      ""
    ),
    envir = environment(),
    quiet = TRUE
  )
})
cat(unlist(TC71_resRmd), sep = "\n")

Over-expressed

Enrichr Link: Over-expressed.

Under-expressed

Enrichr Link: Under-expressed.

Discussion

How do we know that the STAG2 KO treatment worked?

What does comparing the SA2 KO samples from both cell lines tell us?

What SA2 KO DEGs are specific to TC71? A673? and which are shared between the two?

Any Notable results?

Works Cited

Adane B, Alexe G, Seong BKA, Lu D, Hwang EE, Hnisz D, Lareau CA, Ross L, Lin S, Dela Cruz FS, Richardson , Weintraub AS, Wang S, Iniguez AB, Dharia NV, Conway AS, Robichaud AL, Tanenbaum B, Krill-Burger JM, Vazquez F, Schenone M, Berman JN, Kung AL, Carr SA, Aryee MJ, Young RA, Crompton BD, Stegmaier K. STAG2 loss rewires oncogenic and developmental programs to promote metastasis in Ewing sarcoma. Cancer Cell. 2021 Jun 14;39(6):827-844.e10. doi: 10.1016/j.ccell.2021.05.007. PMID: 34129824; PMCID: PMC8378827.

Li G, Zhang P, Zhang W, Lei Z, He J, Meng J, Di T, Yan W. Identification of key genes and pathways in Ewing’s sarcoma patients associated with metastasis and poor prognosis. Onco Targets Ther. 2019 May 27;12:4153-4165. doi: 10.2147/OTT.S195675. PMID: 31213834; PMCID: PMC6549663.

Surdez D, Zaidi S, Grossetête S, Laud-Duval K, Ferre AS, Mous L, Vourc’h T, Tirode F, Pierron G, Raynal V, Baulande S, Brunet E, Hill V, Delattre O. STAG2 mutations alter CTCF-anchored loop extrusion, reduce cis-regulatory interactions and EWSR1-FLI1 activity in Ewing sarcoma. Cancer Cell. 2021 Jun 14;39(6):810-826.e9. doi: 10.1016/j.ccell.2021.04.001. Epub 2021 Apr 29. PMID: 33930311.

Tirode F, Surdez D, Ma X, Parker M, Le Deley MC, Bahrami A, Zhang Z, Lapouble E, Grossetête-Lalami S, Rusch M, Reynaud S, Rio-Frio T, Hedlund E, Wu G, Chen X, Pierron G, Oberlin O, Zaidi S, Lemmon G, Gupta P, Vadodaria B, Easton J, Gut M, Ding L, Mardis ER, Wilson RK, Shurtleff S, Laurence V, Michon J, Marec-Bérard P, Gut I, Downing J, Dyer M, Zhang J, Delattre O; St. Jude Children’s Research Hospital–Washington University Pediatric Cancer Genome Project and the International Cancer Genome Consortium. Genomic landscape of Ewing sarcoma defines an aggressive subtype with co-association of STAG2 and TP53 mutations. Cancer Discov. 2014 Nov;4(11):1342-53. doi: 10.1158/2159-8290.CD-14-0622. Epub 2014 Sep 15. PMID: 25223734; PMCID: PMC4264969.

Yan C, Wang Y, Wang Q, Feng X, Wang L, Bu Z, Lu B, Jiang J. Identification of key genes and pathways in Ewing’s sarcoma using bioinformatics analysis. J BUON. 2018 Sep-Oct;23(5):1472-1480. PMID: 30570875.